Platform Motivation
The widespread adoption of sustainable biobased nanocomposites faces several key challenges, including (1) the lack of comprehensive, high-quality experimental datasets, (2) inefficient dissemination mechanisms that limits collaboration among diverse stakeholders, and (3) absence of accessible data platforms and user-friendly visualization tools. To address these persistent challenges, we have established a data-sharing platform that compiles approximately 1 billion formulations of biobased nanocomposites, along with their machine learning (ML)-predicted properties. This platform is part of a research project led by Dr. Po-Yen Chen’s group at the University of Maryland, College Park.
Platform Description
This data-sharing platform features three key functionalities forward prediction, inverse design, and life cycle assessment (LCA) analysis.
In the forward prediction tab, users can select a set of any four or five components at varying ratios. The platform then uses its embedded prediction model to forecast the optical, fire-retardant, and mechanical properties of a biobased nanocomposite from the selected formulation.
In the inverse design tab, users can specify target property requirements, prompting the platform to perform cluster analyses using the embedded ML-enabled prediction model. The platform then reccomends the most suitable formulations, enabling users to interactively optimize compositions within the recommended loading range for each selected component.
In the LCA analysis tab, users can select any combination of four or five components at varying ratios. The platform then calculates the estimated environmental impacts associated with the selected biobased nanocomposite formulation across 10 different categories and compares them with those of commonly used plastic films, including LDPE, HDPE, and PVC.
In the forward prediction tab, users can select a set of any four or five components at varying ratios. The platform then uses its embedded prediction model to forecast the optical, fire-retardant, and mechanical properties of a biobased nanocomposite from the selected formulation.
In the inverse design tab, users can specify target property requirements, prompting the platform to perform cluster analyses using the embedded ML-enabled prediction model. The platform then reccomends the most suitable formulations, enabling users to interactively optimize compositions within the recommended loading range for each selected component.
In the LCA analysis tab, users can select any combination of four or five components at varying ratios. The platform then calculates the estimated environmental impacts associated with the selected biobased nanocomposite formulation across 10 different categories and compares them with those of commonly used plastic films, including LDPE, HDPE, and PVC.
-Hayden Whitley, Tianle Chen, Dr. Yang Li, Dr. Po-Yen Chen